Structural Break

What is structural break?

Structural break is an unexpected change in the pattern of data that we are given to work with.

Types of structural breaks

  1. Cyclic/seasonal — type of structural break where there are repeated patterns in the structural breaks
  2. Non Cyclic — type of structural break where there are no repeated patterns in the structural breaks
Non cyclic structural break
Cyclic structural break

How to detect structural breaks?

We can detect structural breaks by looking at the scatter plot if possible or else,

To detect structural breaks, we can use Chow test

Let the model be defined as:

Linear regression model in matrix form

We can add dummy variables as follows

Linear regression model with dummy variables

Where Delta t are active at time t with feature X

Now we can do a F test to see whether all dummy variables are 0 or not. If all dummy variables are 0 collectively then we can say there is no structural break otherwise there is structural break

Hypothesis of Chow test

If the null hypothesis is accepted then we can conclude that there is no structural break and the data is perfect to work with

Ways to resolve structural breaks

  1. If the structural break is a non cyclic structural break, we can divide the data into different parts according to the structural breaks and calibrate different models according to different values of X
  2. If the structural break is a cyclic structural break, then we can introduce a dummy variable which has a particular value according to a particular period when the structural break is present or it is 0 otherwise

--

--

--

A data science enthusiast currently doing bachelor's degree in data science

Love podcasts or audiobooks? Learn on the go with our new app.

Data Science Immersive and the Bandaid Paradox

How to make a checkerboard, plot it and predict on it using sklearn

Using machine learning to generate recipes that actually works

Federica Fragapane’s Stories Behind the Line

CXL Review: Digital Analytics IX

Visual Product Search for Smart Retail Checkout

An algorithm for caching predictive models

How I won my tenth bronze medal by analysing the Kaggle Titanic dataset

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Aayushmaan Jain

Aayushmaan Jain

A data science enthusiast currently doing bachelor's degree in data science

More from Medium

Statistics for Data Science

Random Variables And Its Types In Statistics In Just 5 Minutes

Off To See the Wizard:

How Data Science is helping to understand brain intricacies: